Chapter 12 :Congestion in Data Networks pdf

William Stallings
Data and Computer
Communications
Chapter 12
Congestion in
Data Networks

What Is Congestion?

Congestion occurs when the number of packets
being transmitted through the network
approaches the packet handling capacity of the
network

Congestion control aims to keep number of
packets below level at which performance falls
off dramatically

Data network is a network of queues

Generally 80% utilization is critical

Finite queues mean data may be lost

Queues at a Node

Effects of Congestion

Packets arriving are stored at input buffers


Routing decision made

Packet moves to output buffer

Packets queued for output transmitted as fast as
possible

Statistical time division multiplexing

If packets arrive to fast to be routed, or to be
output, buffers will fill

Can discard packets

Can use flow control

Can propagate congestion through network

Interaction of Queues

Ideal
Performance

Practical Performance

Ideal assumes infinite buffers and no overhead

Buffers are finite

Overheads occur in exchanging congestion

control messages

Effects of
Congestion -
No Control

Mechanisms for
Congestion Control

Backpressure

If node becomes congested it can slow down or
halt flow of packets from other nodes

May mean that other nodes have to apply
control on incoming packet rates

Propagates back to source

Can restrict to logical connections generating
most traffic

Used in connection oriented that allow hop by
hop congestion control (e.g. X.25)

Not used in ATM nor frame relay

Only recently developed for IP

Choke Packet


Control packet

Generated at congested node

Sent to source node

e.g. ICMP source quench

From router or destination

Source cuts back until no more source quench message

Sent for every discarded packet, or anticipated

Rather crude mechanism

Implicit Congestion Signaling

Transmission delay may increase with
congestion

Packet may be discarded

Source can detect these as implicit indications of
congestion

Useful on connectionless (datagram) networks

e.g. IP based


(TCP includes congestion and flow control - see chapter 17)

Used in frame relay LAPF

Explicit Congestion Signaling

Network alerts end systems of increasing
congestion

End systems take steps to reduce offered load

Backwards

Congestion avoidance in opposite direction to packet
required

Forwards

Congestion avoidance in same direction as packet
required

Categories of Explicit Signaling

Binary

A bit set in a packet indicates congestion

Credit based


Indicates how many packets source may send

Common for end to end flow control

Rate based

Supply explicit data rate limit

e.g. ATM

Traffic Management

Fairness

Quality of service

May want different treatment for different
connections

Reservations

e.g. ATM

Traffic contract between user and network

Congestion Control in Packet
Switched Networks

Send control packet to some or all source nodes


Requires additional traffic during congestion

Rely on routing information

May react too quickly

End to end probe packets

Adds to overhead

Add congestion info to packets as they cross
nodes

Either backwards or forwards

ATM Traffic Management

High speed, small cell size, limited overhead bits

Still evolving

Requirements

Majority of traffic not amenable to flow control

Feedback slow due to reduced transmission time
compared with propagation delay

Wide range of application demands


Different traffic patterns

Different network services

High speed switching and transmission increases
volatility

Latency/Speed Effects

ATM 150Mbps

~2.8x10
-6
seconds to insert single cell

Time to traverse network depends on
propagation delay, switching delay

Assume propagation at two-thirds speed of light

If source and destination on opposite sides of
USA, propagation time ~ 48x10
-3
seconds

Given implicit congestion control, by the time
dropped cell notification has reached source,
7.2x10
6
bits have been transmitted


So, this is not a good strategy for ATM

Cell Delay Variation

For ATM voice/video, data is a stream of cells

Delay across network must be short

Rate of delivery must be constant

There will always be some variation in transit

Delay cell delivery to application so that
constant bit rate can be maintained to
application

Time Re-assembly of CBR Cells

Network Contribution to
Cell Delay Variation

Packet switched networks

Queuing delays

Routing decision time

Frame relay


As above but to lesser extent

ATM

Less than frame relay

ATM protocol designed to minimize processing
overheads at switches

ATM switches have very high throughput

Only noticeable delay is from congestion

Must not accept load that causes congestion

Cell Delay Variation
At The UNI

Application produces data at fixed rate

Processing at three layers of ATM causes delay

Interleaving cells from different connections

Operation and maintenance cell interleaving

If using synchronous digital hierarchy frames, these
are inserted at physical layer

Can not predict these delays

Origins of Cell Delay Variation

Traffic and Congestion
Control Framework

ATM layer traffic and congestion control should
support QoS classes for all foreseeable network
services

Should not rely on AAL protocols that are
network specific, nor higher level application
specific protocols

Should minimize network and end to end system
complexity

Timings Considered

Cell insertion time

Round trip propagation time

Connection duration

Long term

Determine whether a given new connection can
be accommodated


Agree performance parameters with subscriber